$0b market jensen quote

Z / Runner 🏃🏻‍♂️

🏃🏻‍♂️ worthwhile to have its own bullet at the top. Jensen has said so many things, but this one hit me the hardest… $0B market (it removes so much love, envy, lust, doubts, joy, pain, inspiration…) when i think about startups and opportunities. If it is a $10B market, then I am probably already too late.

🏃🏻‍♂️ testing super ninja: in terms of basic tooling, comparable w/ manus, but lacks the depth & complexity of manus (virtue machine spinup is slower), reasoning feels similar. Need to upgrade to paid account for more robust testing.

• 🏃🏻‍♂️ oh no, was hoping for a team of chinese / russian engs w/o any "moral" shackles /jk

🏃🏻‍♂️ re: healthbench is medical-grade/clinical (lifespan), not lifestyle/behavioral (healthspan). It most likely will be x% better than most doctors in terms of reasoning, but won't have coach-like intervention. Not a weakness, it is a scope clarity.

• 🏃🏻‍♂️ tbd, apple heath is the only interesting one b/c it quickly bootstraps their sensor data integration/aggregation. The bigger question remains, will ppl use chatgpt as a lifestyle & healthspan? The current 230 million success stories maybe due to lack of viable alternatives… i suspect, it will not play out like vertical search w/ google, which a very few were able to compete. mental construct: did google win the travel search/booking space after 20 years?

• 🏃🏻‍♂️ factually inaccurate, but again, depends on your define it (airbnb, expedia, bookings… just to throw some names).

• 🏃🏻‍♂️ confused about how you define vertical search… you mean like search "jobs", "restaurants", "food delivery"... extremely granular?

• geo: 40 countries, split across east/west, openai scale precludes hyper-localization. They can't reason at the city-level (food, fitness, social norms… community level patterns). It is a good place for startup to outperform.

• 🏃🏻‍♂️ geo not in the sense of 4square, more like waze:google maps.

• 🏃🏻‍♂️ simple example, google can't index what local farmers are selling at the local supermarkets… another example (at scale), weather predictions are run/managed as a distributed network (hate to use this term, crypto context), effectively, the sensor towers/stations are managed by small scale companies and the data aggregates / sold to large scale prediction models (weather channel)

• 🏃🏻‍♂️ the food indexing pattern is stuck at circa 2000s, in this instance, you want gmail/visa to index actual food purchases / consumption. The hard problem you tackled a decade ago is a solved problem (google, yelp…), the last mile problem. Your intuition was right, "show me the receipt of your costco". Google isn't tackling that market yet, b/c it is $0 billion market. They want to the local ads business, not the what's on my plate business (yet).

• 🏃🏻‍♂️ what about "eating" agents? I don't mean that at the surface level, i mean what goes into my mouth, how & when… black box for all (status quo). Only my wife knows, and she only knows <50%. The easy & low hanging fruits are gone (90s, 2000s…), but the hard & high hanging fruits remain. Just like google didn't bother to index traffic at the car level, so waze did it for $1.5b back in 2013. Hope this doesn't dove tell into a solomo (food), my point is there are still many "dark corners" of the day to day info yet to be digitized. And like all previous problems, it seems hard, boring, not scaleable, not valuable… until someone solves it. The bigger point is longevity is locally dependent, that's why blue zones book highlights unique and geo specific properties.

• 🏃🏻‍♂️ you mean fx health and BJ are all doomed? Plus, maybe your definition / scope of longevity maybe different.

• 🏃🏻‍♂️ don't aggregate, intervention & create a forcing function to "do" something… dumb example, BJ didn't just tell us how to live longer, he is telling/selling us stuff to take. knowing/understanding longevity (what we did in 2025) != doing longevity. To be fair, not everyone has the time/motivation to play bball or run miles, or pay for healthy grocery… so many hard problems to solve.

• Nagging question: will openai (and many frontier model companies) successfully morph from a de-facto productivity platform into other market categories w/o breaking company DNA. history suggests this is hard, google tried social & failed, b/c it was & will always be an information company. ChatGPT & many ai model companies are fundamentally general intelligence focused.

• 🏃🏻‍♂️ i see the same thing but, prediction != reality. FB failed at hardware for nearly 2 decades, apple sucks at cloud… i don't have the answer, but historical patterns are murky at best.

• 🏃🏻‍♂️ I don't think function health integration is the key, the general agent for workflow is a much deeper and valuable approach, hence my comment yesterday was about the end-state and where will the agent converge w/ the application. We both realize that aggregating data is not a business. I don't think openai is in the longevity space, as indicated by their paper (very focused on medical grade research)... my prediciton is that most ppl will have chatgpt fatigue, just like i don't want to use google for social and fb for productivity. Sometimes, having that separation is the natural order of things. I always say, tencent is the only company that can do both utility (wechat ½ of rev) & fun (game 50% rev)... few large scale companies can do both.

• 🏃🏻‍♂️ definitely not working on yet another chatbot… need to remind self on tackling ideas/problems that exist in the real world (punch thru the 4th wall). Most ppl are not in chatgpt 24/7 like we are (i had more than 2700 chats in 2025), my sister had less than 200, just to show how internal day/day may warp reality. Self-audit, do most really care about living healthier? If so, then the problem of healthspan becomes real. So far, i haven't hear ppl telling me otherwise, so will take it at face value and stay focused.

• 🏃🏻‍♂️ so someone like shamir signing up for $700 biopanel tests, kush doing supplements, aaron buying whoop (even though he said, it is a wasted of wrist), sister signing up for gym / taking creatine… these don't count? I think you mean paying for digital right?

• 🏃🏻‍♂️ Side note: ai(j) and any other ai related business should be subscription based like the old saas market… it should be outcomes based pricing or negative commitment-contract.

• 🏃🏻‍♂️ naw… you may be right and i could be seeing the colors through my own "filters", but that's the conviction… almost year of living health, and still feel like i just scratched the surface… and knowing more and more ppl are practing/struggling… means it is a latent problem. Plus, by design, it is an "unsolvable problem… it takes day/day… into decades of doing.

• 🏃🏻‍♂️ That's why zuck said, the platform may change, but he will always stick w/ social and he referenced fb as a "fashion" problem. It never gets solved, just more problem creating more opportunities.

• My sentiment this week, Healthbench validates my thesis, we have intelligence overhang, operational level decision making & acting per constraints is the real gap. Openai is doing that.

• 🏃🏻‍♂️ reading their paper, they didin't have new technical breakthru as you asked, the reasoning is around O3 model level (2024 sota), what they did was "fine-tune" and benchmark on how real doctors interact w/ real patients. Less about the new model knowing more medical, but being able to communicate (answer) questions more accurately, completely… hence the new rubrics they developed.

• 🏃🏻‍♂️ Side note: this is why i said openai is a much better product company than google, GDM will do crazy new protein folding but won't bother to commercialize it directly w/ big pharma.

• My hope (not a strategy): startups still have room to own the loop due to: grayzone, latent intents, loss/reward incentives, real world behavior.

• 🏃🏻‍♂️ big company can't form pseudo "client-patient" relationship, most likely want to avoid the perception of an "ai coach" in the medical/health context, lawsuit waiting to happen. They already got in trouble when a kid committed sucicde (so did character.ai).

• 🏃🏻‍♂️ loss/reward, in my current ai(j) design i have market/monetary hooks among peers, i don't think openai will tackle this… real world behaviors, strava acquired an ai running coach startup (runna), b/c it uses the ai to push/pull more users into various running events.

• 🏃🏻‍♂️ agree, have you defined/scoped GA (other than what i outlined per manus)?

• 🏃🏻‍♂️ the book is really dense… just listening to the notebookLM summary feels like the knowledge is "locked up".

• 🏃🏻‍♂️ just like the comment about optimal shooting angles, so much of knowledge is not "actuatable". I know i need to sleep, but until i start to measured and care, i didn't matter how many books i read, or mom/wife telling me i need more… only the lived-thru experience and "mental-attention" on the numbers moved me.

• 🏃🏻‍♂️ for the record, your mind maybe built differently, i like to automate my training as much as possible, so sleeping, recovery, calories… are tracked/calculated for me. I suspect most ppl are like me… you don't draw motivation from numbers, but most use it to gauge their progress and use it as a warning sign.

Stephen / Basketball 🏀

• 🏀 don't (die) as an artisit or philosopher.. definitely not analyst!

• 🏀 ceo babak pahlavan, 11 years at google, via CleverSense (alfred!) acquisition.

• 🏀 indeed, no match for os / browser / model builder (at manus).

• 🏀 nah, seems very consumer and lifestyle ready, per integrations with: apple health, my fitness pal, weight watchers, alltrails, instacart, peloton, medical records.

• 🏀 yes, many $100m travel companies but not $10b.

• 🏀 fair, airbnb has 2.6b profit (but not as vertical search).

• 🏀 every site has search; some even started as aggregator. monster(.com) is dead; indeed(.com) also 2.6b profit.

• 🏀 geo startups had all died horribly.

• 🏀 1% differentiation. what local-specific food have impact on longevity? gemini can reason about local shops / customs pretty well by now.

• 🏀 i researched food sources extensively a few weeks ago; gemini was very good at indexing and sourcing all already. see "berkeley bowl" below.

• 🏀 yes, shopping agents are valuable; hence openai's integration with instacart.

• 🏀 nah, all vertical searches had died horribly; ditto for all maps startup – except waze (and mapbox?). i can easily bet that no longevity ai / agent startup will gain $10m revenue in 2026.

• 🏀 function health is / will be doing great. i meant aggregation for longevity via ai / agent will not be competitive.

• 🏀 hence, my second point yesterday: simple (already free) rules.

• 🏀 per peak's podcast: all model companies are launching and capturing well application-level intelligence.

• 🏀 any predictions on the negative (openai sucks at longetivity) are probably not helpful for us; its launch proves the closing of gaps: we can't just do reasoning agent + function health integration.

• 🏀 i don't find the distinction (utility vs fun, or longevity vs healthspan, or medical vs lifestyle) to be critical. per peak's podcast, chatbot war may already be won – except via google/microsoft distributions, definitely not for startups.

• 🏀 nah, that was my key question for you yesterday: no one will or should pay $20 for longevity. let's design an experiment to settle that with 5 people in 1 week. (disprove yourself / ourselves.)

• 🏀 long live tokens / incentives / mechanism designs – nah, just $40/month.

• 🏀 your journaling / ai in particular…

• 🏀 but these points are 100% right and i fully agreed. hence, my third point yesterday: longevity is great focus of lifestyle, but not the most _ product to build company around; definitely not ai.

• 🏀 sans victor bias. web1 stories / web2 analysis / web3 motivations are not what you and i need now.

• 🏀 not sure how this narratives help you / us. the models (with web search) are definitely getting insanely better, but not necessarily opening up for startups to capture as new product launches.

• 🏀 indeed rubrics are great.. once you have 1b users.

• 🏀 true, google docs / maps / _ all suck, but the underlying tech is so solid for 2b users.

• 🏀 what gray zones? which latent intents? why loss/reward incentives? how real-world behaviors?

• 🏀 probably you don't want that (legally binding relationship) either.

• 🏀 because it's (tokens / incentives) too niche; let's think general agents!

• 🏀 that's exactly what i want both of us to deep dive / prototype / discuss 100% time.

🏀 mark's partner stretching: contract-and-relax, karel lewit's post-isometric relaxation (pir) [travell and simons], proprioceptive neuromuscular facilitation (pnf).

• 🏀 indeed very medical grade and industry secrets. mark is ready to spread his knowledge, but very limited bandwidth and incentives. but partner stretching is perfect for social / fit^3! i will try to record my session (as audio) or get some similar videos out of mark.
• 🏀 "on the numbers" means wearables and measuring? i gave up / took off my whoop.
• 🏀 ha, most runners vs _.